Loading ToolBench
Model Context Protocol服务器，用于抓取微博用户信息、动态和搜索等功能
Server has 5 tools with basic Zod schemas and Chinese descriptions. All tools are READ_ONLY and follow verb_noun naming (search_*, get_*). However, descriptions are minimal (10-30 chars), lack context on WHEN to use each tool, and provide no guidance on tool composition or dependencies. Schemas are present but sparse, no enums, no min/max constraints, no output schema documentation. Parameters lack depth (e.g., 'limit' has no bounds). Error handling is silent (try/catch returns empty objects). No tool annotations. Output is raw JSON stringification with no field documentation.
获取微博用户的动态(帖子)
获取微博热搜榜
获取微博用户的资料信息
根据关键词搜索微博内容
根据关键词搜索微博用户
Descriptions are too short (10-30 chars) and lack actionable context. E.g., 'get_profile' is '获取微博用户的资料信息' (18 chars). No guidance on WHEN to use vs. similar tools, what fields are returned, or prerequisites.
No output schema documentation. Tools return raw JSON.stringify() of API responses. LLMs cannot plan downstream calls or extract required fields (e.g., user_id from search_users result).
Parameters lack constraints. 'limit' and 'page' have no min/max bounds, LLMs could pass 0, negative, or absurdly large values. No enums for constrained inputs.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 43 | 2026-07-28+ | v2 |
Error handling is silent. All tools catch errors and return empty objects/arrays with only console.error(). LLMs receive no guidance on what failed, why, or what to do next.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are READ_ONLY but this is not declared in the tool definition, forcing LLMs to infer safety from descriptions.